Goodbye sweet prince!
This hasen't worked for quite some time, Webkit uses the curl backend
currently which has no support for gopher.
Our netkit backend also never "truely" supported gopher, instead
translating pages to html with some prebuild styles, those also don't
use the CSS system colors, for lack of existence back then.
Change-Id: I5d5098db9512bb22cde8c17fa1553637b30ae65b
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8938
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Since BObjectList is a template class, this only breaks ABI where
BObjectList was exposed in public methods, and even then it's only
a name mangling break and we should be able to add compatibility
methods if necessary.
(The old "bool owning" member variable is left intact for ABI
compatibility, for the moment, though it's otherwise unused now.)
Tracker's PoseList is the only remaining type that has a "bool owning"
switch in the constructor rather than template parameters.
This should significantly improve the output of static code analysis
tools that previously detected list operations as causing use-after-frees
and double-frees, as well as make code maintenance easier by making it
easier to determine what list owns (or does not own) an object.
It should also be a minor performance optimization, since the branches
for calls to delete/free should now be optimized out altogether.
Still boots to desktop and Tracker, Deskbar, Debugger all tested
and verified as working.
Change-Id: If2a24a6f0d22e7a506ef554fcfdd328907279ed4
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8915
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
Simplifies the API and makes ownership semantics clearer.
Change-Id: I59d2e3ef8055253e85943500fb65356762055c54
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8914
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
This is to make BMenuBar and BMenu margins the same.
The hardcoded menu item margin numbers come from this math on what the
BMenu and BMenuBar item margins respectively used to be a long time ago:
14 - 8 = 6 // for left margin
(14 + 20) - (8 + 8) = 18 // for left + right margin
We have since updated these margins so they need to be recalculated.
Use BPrivate::MenuPrivate to get the new margin deltas.
IconMenuItem and ModelMenuItem classes are used by Tracker add-ons and
those can be in the menu bar as well.
Change-Id: Ie4147c31e6b19764cb36b9c414f7e91c47b7d9bd
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8905
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: Adrien Destugues <pulkomandy@pulkomandy.tk>
to silence libasan on libtracker.so
Change-Id: I7c634271f4ec2b4ca0d5f76c18ddbfc7dddf2793
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8910
Reviewed-by: Axel Dörfler <axeld@pinc-software.de>
Make it clear that this enum just defines ASCII control characters, and
there is no 'free space'. Also add comment showing which key combination
produces each control character.
Change-Id: I01884132842281cce6c13dd9398b1aeddadae58a
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8899
Reviewed-by: nephele nephele <nep-git@packageloss.eu>
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
* Drop all HOST_KERNEL_* definitions, they aren't used or needed
anymore, it seems.
* Use --no-as-needed unconditionally; it works under Clang+LLD now.
(and indeed it's still needed for the "unzip" host tool to
compile and link properly.)
* Drop --copy-dt-needed-entries. It was added around the time of
GCC 10, but it seems that it's not needed anymore, or at least I
did a minimum build having compiled the host tools without it
and all seemed to be fine. Furthermore, it seems lld doesn't support
it, so this fixes the build under FreeBSD and other platforms
that also don't use ld.bfd same as Darwin/macOS.
* use -fstack-protector-strong and -fstack-clash-protection
* also bin and some user add-ons
Change-Id: I593365e55414ac447bde54980a5c73202e9bb172
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8887
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
Reviewed-by: Adrien Destugues <pulkomandy@pulkomandy.tk>
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
Replace private _UsesSystemColors() with public HasSystemColors()
and make sure to consider the tints when checking if system colors
are used or not.
Document HasSytemColors() in the Haiku Book.
Change-Id: I78de4904d5ddb24b98ad27eb93d4e5ccb330d76a
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8897
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
... and set alternates. This is updated because we can now set
shortcuts without B_COMMAND_KEY. Setup Alt for Win/Linux mode.
* Word-wise shortcuts Option+arrows.
* Line-wise shortcuts Alt+arrows AND Ctrl+arrows for
Win/Linux mode and for when there's a conflict.
Otherwise these shortcuts are not expected to conflict.
Add shortcuts for Alt+Backspace and Alt+Delete to delete to
the beginning or end of line instead of word.
Split out vertical and horizontal shortcuts so that if one is
used by an app we at least get the other one.
Fixes #9913
Change-Id: I0124fec7df4585a70ded8d3e7bf2aa8cb4acecb4
Reviewed-on: https://review.haiku-os.org/c/haiku/+/7289
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: John Scipione <jscipione@gmail.com>
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
... fixing a TODO now that #7078 is fixed.
There are 2 remove playlist actions: 'Remove' and 'Move file to Trash'
* 'Remove' playlist shortcut: Cmd+Delete => Delete
(This also works from the main window to remove the file you're
currently focused on/listening to from the playlist.)
* 'Remove file from playlist' shortcut has been removed. We have
decided not to Trash files from MediaPlayer's playlist anymore,
you'll have to trash the files from a regular Tracker window.
The 'Remove' shortcut gets plain old delete as the author intended.
Change-Id: I3b1e19405e029d9c5f398cfd7a8dbab49fb21fc7
Reviewed-on: https://review.haiku-os.org/c/haiku/+/7292
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
Reviewed-by: nephele nephele <nep-git@packageloss.eu>
Reviewed-by: John Scipione <jscipione@gmail.com>
Introduce B_NO_COMMAND_KEY modifier that is used when a
shortcut has no Command modifier, otherwise B_COMMAND_KEY
modifier is added.
This is conceptually different from B_NO_MODIFIERS in that it
allows other modifiers such as B_CONTROL_KEY, B_SHIFT_KEY and
B_OPTION_KEY to be added to a shortcut, but not B_COMMAND_KEY.
This option is required to remove B_COMMAND_KEY because the
Command modifier is assumed to be present on all shortcuts.
This allows us to break this assumption without breaking apps
written before this change and assume B_COMMAND_KEY to be
present even if it has not been explictly added.
Document B_NO_COMMAND_KEY in BMenuItem and BWindow docs.
Fixes #7078
Co-authored-by: X512 <danger_mail@list.ru>
Co-authored-by: Ryan Leavengood <leavengood@gmail.com>
Change-Id: If0e20cd848112898f4425ede8c17231dc50deb8a
Reviewed-on: https://review.haiku-os.org/c/haiku/+/7293
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: nephele nephele <nep-git@packageloss.eu>
* fixed fifo depth may overflow on uint8
* Created pci/acpi device list with its respective platform version. This will allow us to toggle specific platform registers
Change-Id: I6f4f63649eb62fbc9480e97ad0aa16125458ce7e
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8888
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
Reviewed-by: Jérôme Duval <jerome.duval@gmail.com>
* Drop endian.h; no longer needed as FreeBSD provides its own
(and this conflicts with some other things later in the build.)
* Needs libgnuregex; enable and adjust paths.
Doesn't fully work yet. The fixes in the next commit for fat_shell
plus some other build system tweaks are still needed.
Regression from hrev58560.
Originally: `be_app->PostMessage(&message);`
In hrev58560 it was changed to: `Window()->PostMessage(&message);`
Now: `BMessenger(kTrackerSignature).SendMessage(&message);`
The distinction between original and now is for if Tracker is being
used as a kit, in practice this shouldn't make a difference.
hrev58560 was not correct though.
Fixes #19381
Change-Id: Ia48a8adcf2c9853e8309266369ccf7452c20a679
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8890
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
warning: too many arguments for format [-Wformat-extra-args]
Change-Id: If35737dc2fe193923aa1d756cfd10ca6d820a09e
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8892
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
It was initially added by tqh in ac827a2baa5e0c486b554cdee56c72624b499a40,
but was later removed in 39a81e5ac632d3356ebd4c0b40fb24aa3228833d
due to being rather broken.
This new version is loosely based on that design, but removes the
comparison logic out of the loop and lets the unaligned version
deal with the actual per-character comparison.
Saves around 0.1s out of 3.5s in the best case in a "jam -q HaikuDepot"
run with nothing to do (and that's only comparing with different
libroots on the same kernel; as this strcmp is used in the kernel as
well, we will benefit from this optimization there, too.)
* Add a parameter to RequestAllocator::AllocateAddress and ::AllocateData
that allows the client to specify a minimum amount of free space that
must remain in the port buffer.
* Make use of the new parameter in some operations that can fail
without it.
The Port buffer can be used to store data associated with a Request.
For some file system operations, further requests must be sent through
the port (by calling AllocateRequest) after reserving port buffer
space for data. Unlike AllocateAddress and AllocateData, which can
use an area if the data is larger than the port buffer capacity,
AllocateRequest can only allocate space in the port buffer. If data
previously allocated in the port buffer happens to be large enough to
fill it, then these further AllocateRequest calls will fail.
Change-Id: If03e0afdfbd9fbc36f0e1a04b5d0a20031932b91
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8866
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
The GCC "cc1plus" binary is over 40MB, so pre-faulting at most 10MB
of it only helped so much. Now, we map 1 MB for every time the cache
has been faulted "in full". This should warm up quite rapidly during
compile jobs, and stabilize after that (with future faults mostly
being CoW.)
Only active (accessed+used) pages will get pre-mapped anyway, so if
on a long-running system the page daemon debuffs all the pages in a
cache, we would just iterate through all of them but not map any here.
If that overhead proves to be a problem in the future, we can optimize
for it then (probably by keeping track of how many pages are eligible
for pre-mapping in this way.)
This significantly reduces lock contention in compile jobs. Compiling
libroot with a generated directory on ramfs and -j4, best of 3 runs:
before:
real 0m43.017s
user 1m2.811s
sys 0m24.900s
after:
real 0m29.367s
user 1m0.321s
sys 0m13.373s
Change-Id: I441abf43803c8666cdc13c6d007bc04439698b0b
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8874
Reviewed-by: Michael Lotz <mmlr@mlotz.ch>
This solves a TODO for reporting CoW counts in area_info, and paves
the way for adaptive handling of pre-faulting based on how many
faults a cache has handled.
Change-Id: I4ecd7cf46b794c51acac87184fef49ea5ce76743
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8873
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
We don't really need to clear these buffers, as they're basically
always used with user_strlcpy or precisely sized buffers, so there
should be no risk of "leaking" information to userland.
